在社群媒體高度仰賴圖像傳遞資訊的環境中，圖文已成為消費者理解內容、判斷真實性，並形成心理反應的重要依據。隨著生成式人工智慧技術快速發展，AI生成圖文內容逐漸成為重要的視覺內容形式，也使消費者面對AI生成圖文內容與人工創作圖文內容時，可能產生不同的圖片特徵感知與情緒反應。故本研究以S-O-R模型為理論基礎，將圖片特徵中的吸引力、可靠性與專業性作為外部刺激，並結合PAD情緒模型，探討消費者對AI生成與人工創作圖片之感知差異，以及如何透過此歷程進一步影響購買意圖與口碑傳播。研究採網路問卷調查法蒐集資料，共取得1，048份有效問卷，其中AI生成內容組為549份，人工創作內容組499份，並運用結構方程模式進行路徑分析與假設檢定。實證結果發現，AI生成內容組與人工創作內容組皆顯示，圖片吸引力與專業性能顯著提升喚醒，且支配皆能進一步正向影響購買意圖與口碑傳播。差異在於，可靠性僅在人工創作內容組中能顯著影響喚醒；此外，AI生成內容組呈現較完整的情緒轉化歷程，而人工創作內容組則主要透過喚醒直接提升支配，喚醒對愉悅及愉悅對支配之影響皆未達顯著。此結果有助於釐清不同圖片來源在消費者心理反應中的作用差異，並深化圖片特徵、情緒歷程與行為意圖之間關係的理論理解。;In a social media environment that relies heavily on images to convey information， image-text content has become an important basis for consumers to understand content， evaluate authenticity， and form psychological responses. With the rapid development of generative artificial intelligence technologies， AI-generated image-text content has emerged as an important form of visual content， leading consumers to develop different perceptions of image characteristics and emotional responses when exposed to AI-generated and human-created image-text content. Therefore， based on the Stimulus-Organism-Response (S-O-R) model， this study regards image attractiveness， reliability， and expertise as external stimuli and incorporates the Pleasure-Arousal-Dominance (PAD) emotional model to examine consumers’ perceptual differences toward AI-generated and human-created images， as well as how these processes influence purchase intention and word-of-mouth communication. Data were collected through an online questionnaire survey， yielding 1，048 valid responses， including 549 in the AI-generated content group and 499 in the human-created content group. Structural equation modeling was employed for path analysis and hypothesis testing. The empirical results show that， in both groups， image attractiveness and expertise significantly and positively influence arousal， while dominance positively affects purchase intention and word-of-mouth communication. However， reliability significantly influences arousal only in the human-created content group. In addition， the AI-generated content group demonstrates a more complete emotional transformation process， whereas the human-created content group mainly shows that arousal directly enhances dominance， while the effects of arousal on pleasure and pleasure on dominance are not significant. These findings clarify the differential roles of image sources in consumers’ psychological responses and deepen theoretical understanding of the relationships among image characteristics， emotional processes， and behavioral intentions.
